Skimmed periodic trigger events are used as background overlay on top of simulated data.
These are stored in digi format, in bins of McIlwain-L. Input files are selected by bin, events then selected "randomly".
Overlay is done by merging digis for CAL and TKR with a placeholder for ACD as well.
Status - tagged and is ready to be incorporated into a GR tag on a branch.
Plan - run as part of the system tests.
Exercise the code to look for any general issues.
Directly compare before and after.
The problem - how to get the correct calibration constants for the input background digis.
Only *one* event time can be active at a time in Gaudi.
Periodic triggers can be taken at different times corresponding to different calibrations, and the simulation can be done with calibrations corresponding to yet other times.
The scheme only works correctly for MC generation where "ideal" calibration constants are used, which is clearly not useful for generating "real data."
Upgrade in progress
Get the correct calibration constants to the background digi data
Rewrite the background files with their calibrations applied, leaving them in a form that doesn't need "calibration".
Save information in a new set of overlay data classes.
For CAL and ACD, merge the info a the MC stage, before digis are created.
In the CAL, trigger info needs to be generated before digitization
The production of background overlay files will now be a two step process:
Skim jobs to pull out periodic triggers from the data
Conversions jobs which run at least part of the reconstruction on these files, convert to these new TDS classes and output them to new background overlay files.
Tasks
New TDS classes ???
Modify the overlay code to access the new classes. ??
New code to merge the results of ACD and CAL simulation with the overlay info ?
Create PDS (Root) equivalents (new package overlayRootData)
Convert periodic-trigger Digi files to the new format